Permissible Contradiction, Not Imperissible Impeachment: Section 3 of the Criminal Law Procedure Act 1865 Allows the Prosecution to Call Later Evidence Contradicting Its Own Witness Where the Purpose and Effect Are to Prove Facts in Issue

1) Introduction

R v O'Neill, R. (Rev1) [2025] NICA 69 is a Northern Ireland Court of Appeal decision arising from the conviction of Raymond O’Neill for the murder of Jennifer Dornan and arson with intent at 2 Hazel View, Dunmurry on 2 August 2015. The prosecution case was heavily circumstantial (CCTV tracking, clothing, timings, alleged forensic awareness, and post-offence conduct), but was significantly strengthened mid-trial when Suzanne Hazley came forward and gave evidence that she overheard the applicant confess to his nephew (Shane O’Neill) shortly after the killing.

The appeal comprised eight grounds. The Court granted leave only on:

  • Ground 1: whether the prosecution impermissibly impeached its own witness (Shane O’Neill) by calling Hazley;
  • Ground 7: admission of expert imagery analysis evidence, including later-disclosed regulatory material.

The judgment’s principal legal significance lies in its structured approach to section 3 of the Criminal Law Procedure Act 1865 (“Lord Denman’s Act”) and the boundary between (i) impermissible impeachment of a party’s own witness and (ii) permissible contradiction by calling other evidence relevant to issues in the case—even where that contradiction incidentally damages the earlier witness’s credibility. The Court also provided forward-looking guidance on good practice in relation to forensic imagery/video analysis methodology.

2) Summary of the Judgment

The Court held that:

  • The trial judge’s admission of Suzanne Hazley’s evidence did not breach the common law rule or section 3 of the 1865 Act. The prosecution’s purpose was to prove facts going to guilt (confession and concealment), not to attack Shane O’Neill’s general credibility; and, viewed in the round, the effect of Hazley’s evidence was likewise to fortify the case against the applicant rather than amount to a forbidden credibility assault.
  • The prosecution’s closing address and the judge’s summing up stayed within the permissible boundaries: the prosecution may ask the jury to accept some parts of a witness’s evidence and reject others, without explicitly asserting the witness is deliberately lying unless hostile treatment is obtained.
  • The trial judge made no error in excluding speculative “third party” material (the “army captain” point) as irrelevant hearsay, nor in the various bad character rulings, nor in admitting forensic gait analysis.
  • The imagery expert (Matthew Stephens) was properly admitted; later-disclosed concerns arising from unrelated correspondence between the Forensic Science Regulator and another analyst (Mr Buxton) did not render the conviction unsafe. However, the Court stated that in future it would be preferable for experts in Northern Ireland to be more candid about methodology, the FIAD scale, the Regulator’s position, and accreditation.
  • Applying the safety test in R v Pollock [2004] NICA 34, the conviction was safe; the appeal was dismissed.

3) Analysis

3.1 Precedents Cited

A. The “impeaching one’s own witness” line of authority (section 3 of the 1865 Act)

The Court anchored its analysis in the text of section 3 of the Criminal Law Procedure Act 1865, extracting four rules (impermissibility of general bad character impeachment; permissibility of contradiction where adverse/hostile; permissibility—by leave—of proof of prior inconsistent statement; and the procedural prerequisite of confronting the witness with the circumstances of the statement). Against that statutory architecture, it traced a doctrinal path:

  • EweR v Ambrose [1825] 3 B&C 745: treated as an early articulation of the distinction between (i) forbidden general discrediting and (ii) permitted contradiction by other evidence showing mistake as to a fact. The Court noted the lack of a purpose/effect distinction in the language of the judgments, but emphasised the clear separation between contradiction on the issue and a general assault on credibility.
  • Greenough v Eccles [1859] 5 CB (MS) 786: central for construing “adverse” as meaning “hostile” rather than merely “unfavourable,” and for recognising that the statute introduced a new mechanism (proof of inconsistent statement) in addition to the common law right to contradict by other relevant evidence.
  • R v Prefas (unreported, 11 November 1986) and R v Pacey [1994] (Lexis Citation 3316): treated as limited assistance (fact-specific; section 3 not materially engaged). The Court cautioned against extracting broad statements (“must accept the credit”) as comprehensive statements of law.
  • R v Cairns [2003] 1 WLR 796: used to support the proposition that, in the interests of justice, the prosecution may rely on parts of a witness’s evidence and contradict other parts by calling other evidence, without necessarily proceeding by hostile witness mechanisms. The Court highlighted Cairns’ alertness to the distinction between attacking a witness’s credit and contradicting portions of evidence.
  • R v Jobe [2004] EWCA Crim 3155: primarily a hostile witness case; cited for the practical proposition that an omission may constitute an “inconsistency” in appropriate circumstances. This mattered in the background because the Shane O’Neill “contradiction” in O’Neill arose from critical omissions, though not via a prior inconsistent statement gateway.
  • R v Ross [2007] EWCA Crim 1457: a cautionary example where the “sole purpose” of adducing convictions was to discredit a witness’s forthcoming testimony. Ross served as the paradigm of what section 3(“general evidence of bad character” to impeach credit) forbids.
  • R v Clarke [2011] EWCA Crim 407: relied upon to mark the boundary in closing speeches: prosecution may invite the jury to prefer one body of evidence over another, but must not claim its own witness is deliberately lying unless the point has been put and hostile treatment obtained. The Court distinguished Clarke on its facts but used it as a “do not cross” sign when assessing the prosecutor’s closing here.
  • R v Smith and Others [2019] EWCA Crim 1151: endorsed by the Court; used as the clearest modern statement that the prosecution may call a witness only part of whose evidence it considers true, may adduce other contradictory evidence, and may invite the jury to reject the inaccurate part—without hostile designation—provided it does not introduce prior inconsistent statements as contradiction absent hostility.

B. Fairness, trial management, and public interest

  • Attorney General's Reference No 3 of 1999 [2001] 2 AC 91: Lord Steyn’s “triangulation of interests” principle (accused, victim/family, public) was used as an interpretive “prism” to evaluate whether a technical evidential objection should exclude highly probative evidence and to justify appellate restraint where fairness is preserved.
  • R v H and C [2004] UKHL 3: invoked for Lord Bingham’s “minister of justice” reminder when assessing the propriety of prosecuting counsel’s closing speech.
  • R v Joyce [2023] NI 67: cited for the test for hostility (“unwillingness to tell the truth”/“hostile animus”), illustrating why the “hostile” route was neither apt nor necessary in the circumstances.

C. Third-party perpetrator relevance and speculation

  • R v Greenwood [2004] EWCA Crim 1388: relied on by the defence to justify adducing material pointing to other possible perpetrators. The Court upheld the trial judge’s careful distinction: Greenwood admits evidence with real logical traction; it does not license speculative narratives unsupported by evidence (here, the “army captain” employment detail).

D. Bad character of non-defendants (witness credibility gateway)

  • R v Brewster [2010] EWCA Crim 1194: applied to the Article 5(1)(b) gateway; especially the “significant hurdle” that creditworthiness must be of substantial importance in the case as a whole, and that probative value depends on nature/number/age of convictions. The Court accepted the trial judge’s view that the witness McCaughley was not of substantial importance.

E. Expert imagery/video analysis and evaluative scales

  • A-G's Ref (No. 2 of 2002) [2002] EWCA Crim 2373, [2003] 1 Cr App R 21 (321): used to classify admissible identification-by-image scenarios, including expert comparison evidence when it assists the jury.
  • R v Atkins and Atkins [2010] 1 Cr App R 8: central to approving experience-based evaluative opinions and verbal “support scales” despite no statistical database, and to recommending avoidance of numerical scoring.
  • R v T [2011] Cr App R 9: cited to show that regulator-endorsed methods do not bind the courts, and to reaffirm Atkins’ tolerance of evaluative opinion evidence.
  • R v McDaid [2014] NICA 1: Northern Ireland authority endorsing the FIAD scale in facial mapping, provided the jury is told the opinion is not database-based.
  • R v Abdi [2022] EWCA Crim 315: referenced as a modern example of continued uncriticised FIAD scale use in England and Wales.

F. Post-trial disclosure and safety

  • R v A [2017] NICA 68 adopting R v Hadley and others [2006] EWCA Crim 2544: used to structure the two-question analysis for late disclosure (should it have been disclosed? did non-disclosure render convictions unsafe?).
  • R v Asiedu [2015] 2 Cr App R 8: cited for the approach where late disclosure is remedied on appeal by applying the disclosure test when material comes to light, focusing on whether it truly undermines the conclusions in issue.

G. The appellate safety test

  • R v Pollock [2004] NICA 34: provided the overarching question—whether the conviction is safe.

3.2 Legal Reasoning

A. Ground 1: Contradiction versus impeachment under section 3 of the 1865 Act

The Court’s reasoning is best understood as a disciplined separation of (i) the object of calling evidence and (ii) the collateral consequence that it may undermine a prior witness. The Court accepted that Hazley’s evidence “almost certainly cast a shadow” over Shane O’Neill—because her account suggested he omitted a “second chapter.” But the key legal question was whether this crossed into forbidden territory: impeaching a party’s own witness by general evidence of bad character.

The Court introduced a practical framework:

  • Two situations:
    1. Predictive admissibility rulings (as here): the judge must decide before the witness testifies; “purpose” necessarily dominates.
    2. Post-evidence management: once evidence is in, “effect” normally dominates (directions, possible revisiting of admissibility).
  • Purpose/effect dichotomy: While both may be debated on appeal, in the “ordinary run” of post-conviction appeals the Court stated that effect will be the more important consideration, and the appellate court has an advantage because it can scrutinise what actually happened (transcripts and addresses).

Applying that framework, the Court found:

  • Purpose (predictive stage): the prosecution’s purpose was to adduce highly material facts bearing on guilt (confession; attempts to conceal evidence), not to attack Shane O’Neill’s creditworthiness. Any undermining of O’Neill was incidental.
  • Effect (as the trial unfolded): Hazley’s evidence functioned as direct incrimination of the applicant; it did not become a generalised credibility prosecution against Shane O’Neill. The prosecutor’s closing carefully invited acceptance of some parts of O’Neill’s evidence and rejection of other parts without deploying forbidden “liar/unreliable” language.
  • Hostility route not required: there was no prior inconsistent statement of Shane O’Neill to be proved; the contradiction arose from omissions revealed only once Hazley came forward after his evidence. Designation of hostility (a high hurdle per R v Joyce [2023] NI 67) was not a precondition to calling Hazley.

The Court reinforced its conclusion with an explicit justice-based proposition: it would be “offensive” to exclude “highly material and incriminating evidence” on a “purely technical objection” unconnected to trial fairness; and “no defendant’s trial is rendered unfair by the adduction of material incriminating evidence,” absent a recognised exception.

B. Ground 7: Imagery/video analysis, later regulatory material, and “good practice” guidance

The Court upheld admission of Mr Stephens’ evidence, treating it as a modest but legitimate strand: “limited support” is still “some support,” and the jury could evaluate it, aided by appropriate directions. The Court also took account of the fact that the defence did not instruct a counter-expert.

On the post-trial reliance on material from R v Rainey and others relating to Mr Buxton:

  • The Court rejected the premise that Stephens failed to disclose regulator criticism directed at Buxton. On the evidence, Stephens was likely unaware and had not been personally criticised.
  • Even assuming arguable non-disclosure, the Court applied the R v A [2017] NICA 68/R v Hadley and others [2006] EWCA Crim 2544 approach and concluded that nothing rendered the conviction unsafe.
  • The Court clarified the status of the Forensic Science Regulator in Northern Ireland: persuasive only; the Forensic Science Regulator Act 2021 applies to England and Wales.
  • The Court reaffirmed the acceptability of evaluative verbal scales (FIAD) by reference to R v Atkins and Atkins [2010] 1 Cr App R 8, R v T [2011] Cr App R 9, and R v McDaid [2014] NICA 1.

Importantly, the Court used this ground to issue a forward-looking “lesson”: in future it would be preferable for Northern Ireland experts to explain, as fully as possible, their methodology, the use of the FIAD scale (or alternatives), the FSR position, and the relevance (if any) of ISO17025 accreditation. This is not a change in the admissibility test, but a clear judicial signal about transparency and assisting the tribunal of fact.

3.3 Impact

A. Evidential doctrine and trial practice (section 3 of the 1865 Act)

The decision consolidates a pragmatic, transcript-driven approach to alleged “impeachment”:

  • It reinforces that the prosecution may call later evidence that contradicts an earlier prosecution witness, without automatically triggering hostile witness procedures, where the evidence is directed to facts in issue and not a general bad-character attack.
  • It crystallises the Court’s preference that, in post-conviction review, effect typically matters more than asserted purpose—because the appellate court can examine what actually occurred.
  • It provides a practical template for prosecutors and trial judges navigating the “tightrope”: (i) avoid “deliberately lying” allegations about one’s own witness (cf R v Clarke [2011] EWCA Crim 407); (ii) keep focus on why the contradictory evidence is being called; and (iii) ensure balanced summing up that identifies the conflict and leaves credibility to the jury.

B. Expert evidence in imagery/video analysis

The Court’s “good practice” guidance is likely to have real operational consequences:

  • Experts may increasingly be expected to address the limits of their methodology, the subjective nature of verbal scales, and the absence of population databases.
  • Parties may more routinely litigate (or at least ventilate) accreditation and regulatory guidance, even if not binding.
  • Trial judges may feel fortified in admitting modest “limited support” evidence, provided limitations are clearly explained and properly directed, and the jury has access to the underlying images.

C. Guardrails against speculative third-party narratives

The rejection of the “army captain occupation” point underscores that Greenwood-style third-party perpetrator evidence requires a genuine logical link to facts in issue. Evidence whose real function is to invite the jury to speculate—particularly where disclosure exercises reveal no supporting material—will remain vulnerable to exclusion as irrelevant and/or hearsay.

4) Complex Concepts Simplified

  • “Impeaching your own witness”: a party generally cannot call a witness and then introduce evidence whose purpose is to show that witness is generally untrustworthy (e.g., adducing convictions just to discredit them).
  • “Contradicting your own witness”: a party may call other evidence that conflicts with its witness on a fact in issue; the jury decides which account to accept. This is different from a general credibility attack.
  • Hostile/adverse witness (section 3 of the 1865 Act): if the witness is hostile, the calling party may (with leave) use prior inconsistent statements mechanisms; but hostility is a high threshold and is not automatically engaged whenever evidence is contradicted.
  • PACE Article 76: permits exclusion of prosecution evidence where its admission would adversely affect the fairness of proceedings. The Court emphasised that probative incriminating evidence is not unfair merely because it is damaging.
  • FIAD scale / “limited support”: an expert’s verbal way of expressing how much their comparison findings support one proposition over another. “Limited support” is weak but not zero; weight is for the jury.
  • ISO17025 accreditation: a laboratory/forensic quality standard. Lack of it does not automatically make evidence inadmissible; it may go to weight and transparency, especially where regulator guidance is invoked.
  • Triangulation of interests: courts should keep in view fairness to the accused, the victim/family, and the public interest in effective prosecution—especially when evaluating whether “technical” objections should exclude probative evidence.

5) Conclusion

[2025] NICA 69 is a significant reaffirmation that section 3 of the Criminal Law Procedure Act 1865 does not prevent the Crown from calling later, highly material evidence that contradicts an earlier prosecution witness, where the evidence is directed to facts in issue and not to a generalised attack on the witness’s credibility. The Court’s emphasis on effect (as revealed by the transcript, addresses, and summing up) provides a practical appellate lens for future “impeachment” complaints.

The judgment also signals a modern evidential pragmatism: courts will admit modest strands of expert comparison evidence where it may assist the jury, while encouraging improved transparency about methodology and regulatory context. Ultimately, applying R v Pollock [2004] NICA 34, the Court found the conviction safe and dismissed the appeal.